Is it illegal to use AI to screen resumes?

No, it is not illegal to use AI for resume screening. However, laws like NYC's Local Law 144 require employers to conduct independent bias audits on these tools and publish the results before using them on candidates.

Who is liable if an AI resume screener is biased—the employer or the software vendor?

Under EEOC guidance and Title VII of the Civil Rights Act, the employer is ultimately liable for discriminatory hiring practices, even if the bias originated from a third-party software vendor's algorithm.

What constitutes an Automated Employment Decision Tool (AEDT)?

An AEDT is generally defined as a machine learning, statistical modeling, or AI system used to substantially assist or replace discretionary decision-making in employment, such as scoring resumes, ranking applicants, or conducting automated video interviews.

AI Resume Screening Bias Laws: Employer Compliance Guide 2026

The Rise of Algorithmic Hiring Regulation

The days of deploying an AI tool to rapidly churn through thousands of resumes without legal oversight are over. Recognizing that machine learning models can easily replicate and scale human biases—often penalizing candidates based on gender, race, or age proxies—regulators have established strict rules for Automated Employment Decision Tools (AEDTs).

Leading the charge is New York City's Local Law 144, but federal agencies like the EEOC are actively enforcing existing civil rights laws against modern algorithmic discrimination.

Understanding NYC Local Law 144

If you hire in New York City, you cannot use an AEDT to screen candidates unless you meet three criteria:

1. Independent Bias Audit: The tool must undergo an annual bias audit by an independent auditor.
2. Public Disclosure: A summary of the audit results (including selection rates and impact ratios for protected categories) must be published on the employer's website.
3. Candidate Notice: Candidates must be notified that AI is being used and given instructions on how to request an alternative accommodation.

Vendor Liability vs. Employer Liability

A common misconception among HR leaders is that buying software from a reputable AI vendor insulates the company from discrimination lawsuits. The EEOC has explicitly rejected this defense.

Under Title VII of the Civil Rights Act, the employer is legally responsible for discriminatory hiring outcomes. If an AI resume screener rejects female applicants at a statistically significant higher rate than male applicants (a "disparate impact"), the employer can be sued, regardless of whether the vendor promised the tool was "bias-free."

How Resume AI Becomes Biased

AI models don't need to explicitly look at a candidate's race or gender to discriminate. They find proxies. Historical examples have shown AI models penalizing resumes that included the word "women's" (as in "women's chess club captain") or downgrading graduates from all-women's colleges, simply because the model was trained on historical data overwhelmingly composed of male successful hires.

Compliance Steps for HR Teams

Demand Audit Data: Before signing a contract with an AI screening vendor, demand to see their independent bias audit results.
Monitor Your Own Data: Conduct continuous adverse impact testing on your own hiring funnels (using the 4/5ths rule) to ensure the AI isn't creating disparities in your specific applicant pool.
Provide Accommodations: Ensure candidates have a clear, functional pathway to request that a human review their resume instead of an AI.

Conclusion

AI resume screening offers incredible efficiency, but it carries immense legal risk. By treating AI tools not as infallible judges, but as systems requiring constant auditing and human oversight, employers can leverage the technology while remaining compliant with anti-discrimination laws.